-
ARTICLE
Received 3 Sep 2014 | Accepted 30 Apr 2015 | Published 14 Jul
2015
RC3H1 post-transcriptionally regulates A20 mRNAand modulates the
activity of the IKK/NF-kBpathwayYasuhiro Murakawa1, Michael Hinz2,
Janina Mothes3, Anja Schuetz4,5, Michael Uhl6, Emanuel Wyler1,
Tomoharu Yasuda7, Guido Mastrobuoni8, Caroline C. Friedel9, Lars
Dölken10, Stefan Kempa8,
Marc Schmidt-Supprian11, Nils Blüthgen12,13, Rolf Backofen6,
Udo Heinemann4,14, Jana Wolf3, Claus Scheidereit2
& Markus Landthaler1
The RNA-binding protein RC3H1 (also known as ROQUIN) promotes
TNFa mRNA decay via a
30UTR constitutive decay element (CDE). Here we applied PAR-CLIP
to human RC3H1 to
identify B3,800 mRNA targets with 416,000 binding sites. A large
number of sites are
distinct from the consensus CDE and revealed a
structure-sequence motif with U-rich
sequences embedded in hairpins. RC3H1 binds preferentially
short-lived and DNA damage-
induced mRNAs, indicating a role of this RNA-binding protein in
the post-transcriptional
regulation of the DNA damage response. Intriguingly, RC3H1
affects expression of the NF-kB
pathway regulators such as IkBa and A20. RC3H1 uses ROQ and
Zn-finger domains to
contact a binding site in the A20 30UTR, demonstrating a not yet
recognized mode of RC3H1
binding. Knockdown of RC3H1 resulted in increased A20 protein
expression, thereby
interfering with IkB kinase and NF-kB activities, demonstrating
that RC3H1 can modulate the
activity of the IKK/NF-kB pathway.
DOI: 10.1038/ncomms8367 OPEN
1 RNA Biology and Posttranscriptional Regulation, Berlin
Institute of Medical Systems Biology at the Max-Delbrück Center
for Molecular Medicine, 13125 Berlin,Germany. 2 Signal Transduction
in Tumor Cells, Max-Delbrück Center for Molecular Medicine, 13125
Berlin, Germany. 3 Mathematical Modelling of CellularProcesses,
Max-Delbrück Center for Molecular Medicine, 13125 Berlin, Germany.
4 Macromolecular Structure and Interaction, Max-Delbrück Center
forMolecular Medicine, 13125 Berlin, Germany. 5 Helmholtz Protein
Sample Production Facility, Max Delbrück Center for Molecular
Medicine, 13125 Berlin,Germany. 6 Department of Computer Science
and Centre for Biological Signalling Studies (BIOSS),
Albert-Ludwigs-Universität Freiburg, 79110 Freiburg,Germany. 7
Immune Regulation and Cancer, Max-Delbrück Center for Molecular
Medicine, 13125 Berlin, Germany. 8 Integrative Proteomics and
MetabolomicsPlatform, Berlin Institute of Medical Systems Biology
at the Max-Delbrück Center for Molecular, 13125 Berlin, Germany. 9
Institut für Informatik, Ludwig-Maximilians-Universität, 80333
München, Germany. 10 Institute for Virology and Immunobiology,
University of Würzburg, 97078 Würzburg, Germany.11 Department of
Hematology and Oncology, Technische Universität, 81675 München,
Germany. 12 Institute of Pathology, Charité–Universitätsmedizin
Berlin,10117 Berlin, Germany. 13 Integrative Research Institute
(IRI) for the Life Sciences and Institute for Theoretical Biology,
Humboldt-Universität zu Berlin, 10115Berlin, Germany. 14 Chemistry
and Biochemistry Institute, Freie Universität Berlin, 14195
Berlin, Germany. Correspondence and requests for materials should
beaddressed to M.L. (email: [email protected]).
NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications 1
& 2015 Macmillan Publishers Limited. All rights
reserved.
mailto:[email protected]://www.nature.com/naturecommunications
-
Post-transcriptional regulation of gene expression by
RNA-binding proteins (RBPs) controls a variety of
cellularprocesses. Especially, the modulation of messenger RNA
(mRNA) stability is of critical importance for the
dynamicregulation of genes such as transcription factors and
cytokinesthat need to be switched on and off rapidly1,2.
Roquin is an RBP with a central role in repressing
auto-immunity3. Originally, a missense mutation in the Rc3h1
geneencoding the Roquin-1 protein was identified as the cause
ofsystemic lupus erythematosus-like autoimmunity phenotype
insanroque mice3. Roquin-1 is localized in cytoplasmic
granules4,5
and binds to the 30 untranslated region (30UTR) of
induciblecostimulator (ICOS) mRNA to post-transcriptionally repress
itsexpression6,7. Furthermore, Roquin-1, as well as its
paralogueRoquin-2, interacts with 30UTR of TNFRSF4 and
tumour-necrosisfactor-a (TNFa), and modulates immune responses5,8.
Recentstudies showed that Roquin proteins interact through their
ROQdomains with a constitutive decay element (CDE) in the 30UTR
ofTNFa mRNA and promotes the decay of this transcript byrecruiting
the CCR4-CAF1-NOT deadenylase complex9. The CDEof TNFa folds into a
characteristic stem–loop structure containinga specific
trinucleotide loop, which is highly similar to the Roquin-1 RNA
recognition element in the ICOS 30UTR (ref. 9). Lateststructural
analyses showed the ROQ domain in complex with aprototypical CDE
RNA stem–loop revealing recognition of theRNA stem and its
trinucleotide loop10,11. Leppek et al.9 furtheridentified
additional Rc3h1 target transcripts by RNA-immunoprecipitation
sequencing (RIP-seq) analysis, includingregulators of the nuclear
factor-kB (NF-kB) pathway. However, arecognizable CDE was absent in
the majority of Rc3h1-boundmRNAs, suggesting other modes of RNA
recognition9. In line withthese findings, Schlundt et al.10 showed
by mutational andstructural analyses of RNA ligands that relaxed
CDE consensussequences can mediate Roquin-dependent regulation.
Similarly,Tan et al.11 and Schuetz et al.12 reported that the ROQ
domain ofRc3h1 recognizes the CDE and can also bind to duplex RNA.
Inaddition to the ROQ domain, RC3H1 possesses an N-terminalRING
finger with a potential E3 ubiquitin–ligase function13, as wellas a
CCCH-type zinc (Zn) finger that is involved in RNArecognition7.
CCCH-type Zn-finger RBPs typically contact AU-rich elements14,15.
AU-rich elements are conserved cis-regulatoryelements, originally
discovered in the 30UTRs of short-livedmRNAs, encoding inflammatory
mediators16–18.
To obtain a better understanding of the molecular mechanismsof
human RC3H1 RNA recognition and disentangle its cellularfunction,
we applied PAR-CLIP (photoactivatable ribonucleoside-enhanced
crosslinking and immunoprecipitation)19 to identifyRC3H1
RNA-binding sites and target transcripts in HEK293 cells.RC3H1
contacts mRNAs through structure-sequence elementslocated in
30UTRs. The binding sites are composed of hairpinswith variable
loop length often with embedded U-rich sequences,including CDE
consensus sequences. RC3H1-bound mRNAtargets are short-lived, and
RC3H1 depletion results indecreased mRNA decay rates and increased
protein synthesis ofits target mRNAs. RC3H1 target transcripts are
enriched formRNAs that are induced upon DNA damage, among them
arethe mRNA of A20 (also known as TNFAIP3). A20 codes foran
ubiquitin-editing enzyme, which inhibits activation ofNF-kB20,21.
In vitro and in vivo experiments revealed thatRC3H1 interacts with
a non-CDE-type stem–loop structurepreceded by an AU-rich sequence
in the A20 30UTR involvingROQ and CCCH-type Zn-finger domains,
indicating a yetunrecognized RC3H1-binding mode and specificity.
Depletionof RC3H1 leads to increased A20 protein expression, which
isaccompanied by decreased IkB kinase (IKK) activation andNF-kB
DNA-binding activity upon TNFa signalling.
ResultsHuman RC3H1 binds to thousands of mRNAs. To
identifyRC3H1-binding sites at high resolution, we applied PAR-CLIP
incombination with next-generation sequencing19. In
PAR-CLIPexperiments, nascent RNA is metabolically labelled with the
non-perturbing photoreactive ribonucleosides 4-thiouridine (4SU)
or6-thioguanosine (6SG). Crosslinking of protein to 4SU- or
6SG-labelled RNA leads to specific T to C or G to A
transitions,respectively, that occur at high frequency in
complementary DNA(cDNA) sequence reads and mark the protein
crosslinking siteson the target RNA19. HEK293 cells stably
expressing inducibleFLAG/HA-tagged RC3H1 (Supplementary Fig. 1a)
werecrosslinked after labelling of RNA with either 4SU or
6SG.Immunopurified, ribonuclease-treated and radiolabelled
RC3H1–RNA complexes were separated by SDS–polyacrylamidegel
electrophoresis (PAGE) (Fig. 1a). Protein-protected RNAfragments
were recovered and converted into a cDNA libraryamenable to
Illumina sequencing.
In total, we performed three independent PAR-CLIPexperiments
(two biological replicates with 4SU, 4SU-1 and4SU-2, and one
replicate with 6SG; see Supplementary Table 1).Sequence reads were
mapped to the human genome andoverlapping reads were used to build
RC3H1-binding clusters22.In PAR-CLIP experiments using 4SU,
diagnostic T to Ctransitions detected in mapped reads were most
highlyabundant (Fig. 1b and Supplementary Fig. 1b). Similarly,
butless pronounced, the diagnostic G to A changes were the
mostabundant type of mutation for the 6SG PAR-CLIP
experiment(Supplementary Fig. 1c). A length histogram of RC3H1
PAR-CLIP clusters shows a median cluster size of B25–30
nucleotides(Supplementary Fig. 1d).
We identified B2,000–4,000 RC3H1 mRNA target transcriptsin each
of the 4SU PAR-CLIP experiments (Fig. 1c). Ninety-threeper cent of
the 481 6SG PAR-CLIP mRNA targetswere reproduced in 4SU libraries
(Fig. 1c). We combined the‘reads’ from all PAR-CLIP experiments to
derive a set ofconsensus binding sites supported by reads from at
least twoout of three experiments (see Methods section). Based
onthis analysis, we identified 16,234 RC3H1-binding sites on3,821
protein-coding transcripts as consensus data set(Supplementary Data
1). The binding sites and PAR-CLIPsequence alignments are publicly
available at http://bimsbsta-tic.mdc-berlin.de/landthaler/RC3H1.
The position with thehighest number of PAR-CLIP-derived diagnostic
nucleotidetransitions for each binding sites was referred to as the
preferredcrosslinking site.
To gain an insight into the transcripts regulated by RC3H1,genes
encoding RC3H1-bound mRNAs were subjected to KyotoEncyclopedia of
Genes and Genomes (KEGG) pathway and GeneOntology (GO) term
enrichment analysis23,24. Interestingly, cellcycle and p53
signalling pathway were overrepresented in theKEGG pathway
enrichment analysis (Supplementary Fig. 1e),suggesting that RC3H1
could play a role in the response toDNA damage. Furthermore, GO
term enrichment analysisshowed that RC3H1-bound transcripts are
highly enrichedfor regulators of gene expression such as
transcription factors,RBPs and ubiquitin ligases (Supplementary
Fig. 1f). Inaddition, when comparing the human RC3H1 targets
withmouse Roquin-bound transcripts9, we identified 36 out of
55Roquin-interacting mRNAs by PAR-CLIP in HEK293
cells(Supplementary Fig. 1g).
RC3H1-binding sites are mostly located in 30UTR of mRNAs.Next,
we examined the distribution of RC3H1-binding sites alongmRNA
transcripts. The majority of binding sites (81%) were
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367
2 NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://bimsbstatic.mdc-berlin.de/landthaler/RC3H1http://bimsbstatic.mdc-berlin.de/landthaler/RC3H1http://www.nature.com/naturecommunications
-
found to be located in 30UTRs (Fig. 1d), consistent with
previousobservations that RC3H1 binds to ICOS and TNFa mRNAsthrough
30UTR interactions4,6,7. RC3H1 binding to ICOS andTNFa mRNAs in
HEK293 cells was not observed likely due toundetectable ICOS and
TNFa mRNA expression. A preferencefor RC3H1 binding along 30UTRs of
target transcripts was notapparent, since binding clusters were
almost equally distributedover this transcript region
(Supplementary Fig. 1h). Since aprevious study suggested a
functional link between RC3H1binding and microRNA (miRNA)
activity6, we examined the localinteractions between RC3H1 and
miRNA by computing thedensity of conserved miRNA target sites
around RC3H1 preferredcrosslinking sites (Supplementary Fig. 1i).
The observed profileindicated an overrepresentation of miRNA seed
complements inthe vicinity of RC3H1-binding sites, but did not
directly overlapwith these sites.
U-rich hairpins dominate as recognition features. To
investi-gate RNA features recognized by RC3H1, we searched
forsequence and secondary structure elements in RC3H1-bindingsites.
First, we examined 7mer occurrences in 41 nucleotidewindows centred
on preferred crosslinking sites inRC3H1-binding sites. Notably,
U-rich sequences were frequentlyfound in RC3H1-binding sites
derived from both 4SU and 6SGexperiments (Fig. 2a and Supplementary
Fig. 2a), suggestingthat the frequent observation of U-richness is
not owing to a
bias introduced by using 4SU. These U-rich 7mers
wereoverrepresented in RC3H1-binding sites when compared
withcontrol 7mers (Fig. 2b). In contrast, U-rich elements were
notenriched in IGF2BP1-binding sites19, whereas 7mer containingthe
CAU consensus sequence are overrepresented (Fig. 2b).U-rich
sequences with interspersed adenosines were moreenriched in RC3H1
consensus binding sites in 30UTR sequenceswhen compared with U-rich
7mer sequences containingguanosines (Supplementary Fig. 2b and
Supplementary Data 2).Similar results were also obtained from 5mer
analysis(Supplementary Fig. 2c). In addition, U-rich sequences
werefound in close proximity of preferred crosslink sites,
suggestingthe direct interaction of RC3H1 with these sequences
(Fig. 2c). Inaddition, we examined the occurrence of the previously
identifiedCDE motif9, and found that the core CDE consensus
sequence(UCYRYGA) was present in RC3H1-binding sites, but
thefrequency was less prominent than that of several
U-richsequences (Fig. 2c).
To examine potential secondary structure features in
RC3H1-binding sites, we computationally folded
41-nucleotidesequence stretches centred around the preferred
crosslinkingsites and averaged the resulting base pairing
probabilities.Randomly selected RNA regions of the same length
within30UTRs of RC3H1 mRNA target transcripts served as abackground
control. In 30UTR RC3H1 consensus binding sites,the base pairing
probability was reduced in the vicinity of thecrosslink sites and
increased in the flanking region compared with
250
150
100
75
60
Type of alignment
C:T
C:G
C:A
G:C
G:T
G:A
T:C
T:GT:A
A:C
A:GA:TIns
Del
Edi
tP
erfe
ctT
otal
5.0
2.5
0
4SU-2
4SU-2
4SU-1
c
a b
d
6SG
2,476
1,587
400
232 93
354
RC3H1
×105
5′UTR CDS 3’UTR Intron0
20
40
60
80
Per
cent
age
of to
tal R
C3H
1clu
ster
365 nm6SG4SU – – +
– + –+ + +
Anti-HA
kDa
Num
ber
of m
appe
d re
ads
Figure 1 | PAR-CLIP identifies thousands of human mRNAs directly
bound by RC3H1. (a) Phosphorimage of SDS–PAGE of radiolabelled
FLAG/HA-
RC3H1–RNA complexes from 365 nm ultraviolet light crosslinked
non-labelled, 6SG- or 4SU-labelled cells. Crosslinked protein–RNA
complexes were
observed upon metabolic labelling with 4SU or 6SG. The lower
panel shows an anti-HA western blot, confirming correct size and
equal loading of the IPed
protein. The box indicates the region that was cut out for
PAR-CLIP library preparation. (b) Specific T to C mismatches in
aligned reads demonstrate
efficient mRNA-RBP crosslinking. The frequency of nucleotide
mismatches in 4SU-2 PAR-CLIP reads aligned to mature mRNAs is
shown. Sense mapping is
shown in blue and antisense mapping in red. (c) A Venn diagram
showing the overlap of target mRNA transcripts between 4SU and 6SG
PAR-CLIP
experiments. (d) Distribution of binding sites along mRNA
transcripts based on consensus RC3H1 PAR-CLIP-binding sites. The
majority of binding sites are
located in 30UTRs. CDS, coding sequences.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367 ARTICLE
NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications 3
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
background, suggesting that RC3H1-binding sites tend to
formstem–loop structures (Supplementary Fig. 2d,e). A more
detailedexamination of the types of hairpin structures enriched in
RC3H1binding revealed an overrepresentation of stems capped
bytrinucleotide loops as demonstrated for the CDE motif 9–12,
but
also hairpins with loops containing 4 and 5 nucleotides (Fig.
2d).From this analysis, we concluded that there might be
severalbinding motifs that are likely to be structured. To test
thishypothesis, we applied a variant of an approach that
waspreviously used to detect structured RNA motifs25. Here we
AUUUUUUUUUAAAA
UUUUAUU
UUUUUAAUUUUUUA
UUUUUUU
UUAUUUUUUUAUUU
AUUUUAA
AAAUUUUUAUUUUU
AUUUUAU
AAUUUUU
r (Spearman) = 0.704P < 2.2e–16
−6.0 −5.5 −5.0 −4.5 −4.0 −3.5 −3.0 −2.5
Log10 (7mer frequency - 4SU-2)
−6.0
−5.5
−5.0
−4.5
−4.0
−3.5
−3.0
−2.5
Log1
0 (7
mer
freq
uenc
y -
6SG
)
Crosslink site 14 nt
UUUUUAAUUUUAAAUUUAUUUUUAUUUUUUUUCUUUAUUUAUUCYRYGACUGAACCGGUAUAUUUUUGUU
7mer location in 3’ UTR-binding sites
U-rich with A
Controls0
200
14 nt
CDEARE
U-rich with C
Enr
ichm
ent o
f ind
icat
ed 7
mer
0
1
2
UUUU
UAA
UUUU
AAA
UUUA
UUU
UUAU
UUU
AUUU
UUU
UAUU
UAU
CUGA
ACC
UUUU
GUU
UUUU
UAA
UUUU
AAA
UUUA
UUU
UUAU
UUU
AUUU
UUU
UAUU
UAU
CUUC
AUU
ACAU
UUU
RC3H1-binding sites IGF2BP1-binding sites
U-richwith A
Controls Positivecontrols
0.0
0.5
1.0
1.5
2.0
Enr
ichm
ent o
f ind
icat
edst
em–l
oops
4-3-
45-
3-5
6-3-
67-
3-7
4-4-
45-
4-5
6-4-
67-
4-7
4-5-
45-
5-5
6-5-
67-
5-7
4-3-
45-
3-5
6-3-
67-
3-7
4-4-
45-
4-5
6-4-
67-
4-7
4-5-
45-
5-5
6-5-
67-
5-7
RC3H1 IGF2BP1
GUU
AA A A U A
AA
UAU U U U U
C UU
UUAAAG_A
UU
UAA
AUUUU
AA
AUGUGG
CU
CUGU
UA A A A
UU
UU U C U U U
A _UA
UUAAAGUA
UAU
AU
AU
AU
Motif 1n = 177
Motif 2n = 268
U-richwith A
Figure 2 | Identification of U-rich sequences and stem–loop
secondary structure as recognition elements of RC3H1. (a) Log10
frequencies of 7mers
occurring in the 41-nucleotide (nt) window around the RC3H1
preferred crosslink sites are shown for 4SU-2 PAR-CLIP and 6SG
PAR-CLIP libraries. U-rich
sequences are frequently occurring in both 4SU and 6SG
libraries. (b) Enrichment of indicated 7mers in the 41-nt window
around the RC3H1 (left) or
IGF2BP1 (right) preferred crosslink sites compared with all
30UTR sequences. U-rich sequences with A contents are specifically
enriched in RC3H1 30UTR-
binding sites, whereas 7mers containing known IGF2BP1 motif
(CAU) are enriched in IGF2BP1 30UTR-binding sites to the similar
degree. (c) A heat map
showing the coverage of 7mers, indicated on the left, around the
preferred crosslinks in 30UTR RC3H1 consensus binding sites. U-rich
elements with A
contents and CDE are indicated. U-rich sequences are found in
the close vicinity of crosslink sites, which is indicative of
direct association of RC3H1 with
U-rich sequences. (d) Enrichment of indicated stem–loop
structures in the 41-nt window around the RC3H1 (left) or IGF2BP1
(right) preferred crosslink
sites compared with 41 nt sequences randomly selected from the
30UTRs of target transcripts as a background model. Various
stem–loop structures
(n-m-n indicates a hairpin structure of n-mer stem and m-mer
loop) are enriched in RC3H1 30UTR-binding sites but not in 30UTR
IGF2BP1-binding sites.
(e) The seed alignment and consensus structure of motifs 1 and 2
are shown. ARE, AU-rich elements.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367
4 NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
performed an initial clustering of the set of RNAs, followed by
thedetection of a specific structure. We started with the top
100RC3H1 PAR-CLIP-binding sites, and detected initial clusters
byLocARNA26,27 and RNAclust27. For each subcluster, we usedCMfinder
(version 0.2; ref. 28) to search for a subset of sequencesthat has
a specific sequence-structure motif. CMfinder generatesboth a
sequence-structure alignment (referred to as seedalignment) and a
covariance model, which we used to searchfor further sequences in
the top 1,000 binding sites for remotemembers of this motif using
cmsearch29. The seed alignment ofmotif 1 (present in 177 out of
1,000 binding sites) and motif 2(present in 268 out of 1,000
binding sites), representing the mostfrequently occurring
sequence-structure elements, is shown inFig. 2e. Interestingly, in
motif 2, the preferred crosslinkednucleotides are positioned
upstream of the predicted hairpinstructure, whereas in motif 1 the
preferred crosslinkednucleotides are located in the U-rich loop. In
summary, ourcomputational analyses did not reveal a defined motif.
However,hairpin structures frequently containing U-rich sequences
and,albeit, less frequently, the CDE consensus sequence were
detectedas possible recognition elements of RC3H1.
RC3H1 interacts with CCR4-CAF1-NOT deadenylase complex.To
further investigate the molecular function of RC3H1, we setout to
identify proteins that interact with RC3H1. Previous stu-dies
showed that RC3H1 destabilizes mRNA albeit by differentmolecular
mechanisms. Glasmacher et al.7 reported that RC3H1interacts with
mRNA decapping proteins, whereas Leppek et al.9
more recently showed that roquin protein associates
withCCR4-CAF1-NOT deadenylase complex.
To identify proteins that directly interact with RC3H1 in
anRNA-independent manner, cellular extracts of control
andFLAG/HA-tagged RC3H1-expressing cells were treated
withRNaseT1/I, and the immunoprecipitates were analysed andcompared
by SILAC (stable isotope labelling by amino acids(aa) in cell
culture)-based quantitative mass spectrometry30. As abiological
replicate, we performed a label-swap experiment withreversed
light/heavy isotope labelling (Fig. 3a). As expected,RC3H1 was
efficiently immunoprecipitated as indicated by thelog2
heavy-to-light normalized ratio of 2.80 in the forwardexperiment
and -5.13 in the reverse experiment (SupplementaryFig. 3a). In two
biological replicates, we identified numerouspeptides originating
from components of the CCR4-CAF1-NOT
−1.0 −0.5 0.0 0.5 1.0
0.0
0.2
0.4
0.6
0.8
1.0
Changes in protein synthesis (pSILAC)
Log2-fold change (siRNA/mock)
CD
F
RC3H1 targets Non-targets
IN IP IN IP IN IP
P = 0.0031
−0.4 −0.2 0.0 0.2 0.4
Log2-fold change (siRNA/mock)
0.0
0.2
0.4
0.6
0.8
1.0
CD
F
Changes in mRNA decay rate
RC3H1 targetsNon-targets
P < 2.2e–16
Log2 (number of peptides)
1 2 3 4 50 6
1
2
3
4
5
0
6
Log2
(nu
mbe
r of
pep
tides
) RC3H1
CNOT1
CNOT2
CNOT3
CNOT8
CNOT7
Anti-HA
Anti-myc
QKICNOT8CNOT1
RC3H1RC3H1RC3H1
Figure 3 | RC3H1 recruits deadenylation complex and destabilizes
target mRNAs. (a) A scatter plot of identified peptide counts in
two label-swap
replicates. Peptides eluted from immunopurified FLAG/HA-tagged
RC3H1 complex are analysed by tandem LC-MS/MS. High dose of
RNaseT1/RNaseI are
treated before immunoprecipitation (IP) to disrupt the indirect
interactions mediated by nascent RNA. Peptides derived from the
CCR4-CAF1-CNOT
deadenylase complex were detected. (b) RC3H1 interactions were
confirmed by co-transfection of myc-RC3H1 expression construct with
HA-CNOT1,
HA-CNOT8 or HA-QUAKING (QKI) expression constructs. IP was
performed using anti-myc antibody. IPed proteins were resolved on
SDS–PAGE, blotted
and probed with anti-myc and anti-HA antibodies. Protein
expression in cellular extract used as input for IP experiments is
indicated (IN). (c) A cumulative
distribution function (CDF) plot of log2-fold changes of mRNA
decay rates of the top 500 normalized RC3H1-bound mRNAs is shown in
red and all
expressed mRNAs is shown in black. Top RC3H1-bound mRNAs show
slower mRNA decay rates compared with all mRNAs upon RC3H1/
RC3H2
knockdown. The mean difference in mRNA decay rates (siRNA/mock)
for the top 500 RC3H1 target mRNAs (n¼ 500) and all mRNAs (n¼
15158) are� 1.324 and �0.074, respectively (P value o2.2e–16,
Wilcoxon’s rank sum test). (d) A CDF plot of log2-fold changes of
protein synthesis of consensusRC3H1 target transcripts that have
4100 transitions on 30UTR is shown in red and non-targets is shown
in black after siRNA-mediated RC3H1 depletion.Protein synthesis of
RC3H1-bound mRNAs was upregulated upon RC3H1 knockdown (P value
0.0031, Wilcoxon’s rank sum test). The mean log2-fold
changes for RC3H1 targets (n¼ 390) and non-targets (n¼ 1,279)
are 0.001 and �0.116, respectively.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367 ARTICLE
NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications 5
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
deadenylase complex including CNOT1, CNOT2, CNOT3,CNOT7 and
CNOT8, but not from the decapping complex(Fig. 3a, Supplementary
Fig. 3a and Supplementary Table 2).CNOT1 is the scaffold subunit
and CNOT8 is a catalyticdeadenylase subunit31. Interestingly, for
proteins in theCCR4-CAF1-NOT deadenylase complex, we obtained
SILACratios of B1 suggesting that RC3H1 interacts only
transientlywith these proteins (Supplementary Fig. 3a). Specific
interactionsof RC3H1 with CNOT1 and CNOT8, but not with the RBP
QKI,were confirmed by co-immunoprecipitation experiments
inagreement with previous findings9 (Fig. 3b), indicating thatRC3H1
generally acts as mediator of mRNA deadenylation.
RC3H1 destabilizes target mRNAs. To assess whether
RC3H1interacts with short-lived mRNA transcripts, we performed
tran-scriptome-wide mRNA half-life measurements (SupplementaryData
3) as described by Dölken et al.32, and compared half-lives
ofRC3H1-bound and unbound mRNA transcripts. Consistent withthe
interaction of RC3H1 with the deadenylase complex and apossible
role in mRNA decay, RC3H1-targeted mRNAs werefound to have shorter
half-lives than expressed non-targettranscripts (Supplementary Fig.
3b,c). On the other hand,IGF2BP1-bound transcripts do not show this
tendency(Supplementary Fig. 3d,e). Furthermore, we found that
mRNAhalf-lives of RC3H1-bound transcripts inversely correlated with
anexpression normalized PAR-CLIP score (Supplementary Fig.
3f),suggesting that the extent of RC3H1–mRNA binding determinedthe
mRNA half-lives of bound mRNAs (Supplementary Data 4).
To examine the impact of RC3H1 on the decay rate of itsmRNA
targets, we sequenced mRNA of untreated and RC3H1-and
RC3H2-depleted cells (Supplementary Fig. 3g) after inhibi-tion of
transcription using actinomycin D from two biologicalreplicates.
Figure 3c indicates that mRNA decay rates of RC3H1-bound
transcripts were decreased in cells depleted of RC3H1 andRC3H2,
suggesting that RC3H proteins destabilize their mRNAtargets.
In addition, we examined the effect of RC3H1 depletion on
theprotein synthesis rate of RC3H1-bound mRNAs. For thispurpose, we
monitored changes in newly synthesized proteins bypulsed SILAC
(pSILAC) based quantitative proteomics22,33 upondepletion of
endogenous RC3H1 in HEK293 cells (SupplementaryFig.4a). The mass
shift between the RC3H1 knockdown (‘medium’labelled) and
mock-treated control (‘heavy’ labelled) allowed thequantification
of changes in protein synthesis of B2,400 proteins(Supplementary
Fig. 4a). RC3H1 knockdown was confirmed bywestern blot analysis
(Supplementary Fig. 4b). A cumulativedistribution function analysis
showed that the level of proteinsynthesis of RC3H1-bound mRNAs was
increased upon RC3H1depletion by two short interfering RNA (siRNAs)
(Fig. 3d andSupplementary Fig. 4c,d). Taken together, our data
indicate thatRC3H1-bound mRNAs are short lived, and depletion of
RC3H1resulted in decreased decay rates and increased proteins
synthesisof its target transcripts, validating the functionality of
RC3H1–mRNA interactions.
RC3H1 interacts with DNA damage-induced transcripts. OurKEGG
pathway enrichment analysis revealed that RC3H1 mRNAtargets are
enriched for genes involved in cell cycle regulation andp53
signalling (Supplementary Fig. 1e) Furthermore, RC3H1 wasshown to
localize to stress granules upon oxidative stress inducedby
arsenite exposure4, suggesting an involvement of RC3H1 inthe
cellular stress response. To investigate a possible role ofRC3H1 in
the DNA damage response, we correlated changes inmRNA expression
levels upon DNA damage induced byneocarzinostatin in HEK293 cells34
with our RC3H1 PAR-CLIP
data (Fig. 4a). Notably, RC3H1-bound transcripts were
moreinduced upon DNA damage than unbound mRNAs (Fig. 4b).This
finding implies that the expression levels of DNAdamage-induced
genes are likely modulated at the post-transcriptional level at
least in part by RC3H1. The mRNA ofA20, a NF-kB target gene that
acts as a feedback regulator ofNF-kB activation, was among the
RC3H1 target transcripts andshowed the largest increase in
expression upon DNA damage.Induction of FLAG/HA-tagged RC3H1
(Supplementary Fig. 5a)in HEK293 cells resulted in a significant
reduction of A20 mRNA(Fig. 4c) upon DNA damage. Furthermore, we
observed that A20mRNA half-life was shortened by expression of
tagged RC3H1(Fig. 4d). To provide further support for this finding,
wespecifically blocked the RC3H1-binding site on A20 mRNA
bytransfecting an antisense locked nucleic acid
(LNA)oligonucleotide and observed an increase in A20 mRNA half-life
(Fig. 4e).
RC3H1 binding to A20 30UTR via ROQ and CCCH-Znfdomains. Our
PAR-CLIP data indicated a single RC3H1-bindingsite in the 30UTR of
A20 mRNA. Diagnostic PAR-CLIP T toC transition events, indicating
protein–RNA crosslinking sites,were detected in a loop of a
conserved predicted hairpin and in anAU-rich sequence located
upstream of the stem–loop structure,which differs from the
previously described CDE (Fig. 5a).To examine whether the putative
A20-binding site bestowsRC3H1-dependent mRNA decay, we cloned a
37-bp sequencecovering the crosslinked region into a green
fluorescentprotein (GFP) reporter and assayed mRNA turnover
byquantitative reverse transcription–PCR (qRT–PCR) afterblocking
transcription using actinomycin D. Indeed, insertion ofthe
RC3H1-bound A20 site into the 30UTR of the reporterconstruct
destabilized reporter transcripts in mock-transfectedcells, but not
in RC3H1- and RC3H2-depleted cells (Fig. 5b),indicating that RC3H
proteins destabilize the reporter transcriptsthrough this A20
site.
To further examine the RC3H1 interaction with the putativeA20
site, we used electrophoretic mobility shift assays (EMSA).
Inaddition, to assess the contribution of the different
RC3H1domains to RNA binding, we expressed two variants: RC3H1-N1(aa
2–399) contained the N-terminal RING and ROQ domains,whereas
RC3H1-N2 (aa 2–452) harboured RING, ROQ andCCCH-type Zn-finger
domains (Fig. 5c). Both recombinantproteins bound to the ICOS
CDE-like stem–loop motif RNA(Fig. 5c). The formation of the
protein–RNA complex seemed tobe independent of the CCCH-type
Zn-finger domain. In contrast,the 21-nucleotide A20 hairpin RNA was
bound by RC3H1-N2with higher affinity than RC3H1-N1 (Fig. 5c),
indicating that theCCCH-type Zn-finger domain plays a role in the
interaction withthe non-CDE-type A20 site. The addition of 16
nucleotides 50 ofthe stem–loop structure (A20 37 nucleotide)
further increased theaffinity of the RC3H1-N2 variant to the RNA
substrate (Fig. 5c),suggesting that an additional sequence upstream
of the A20hairpin is involved in protein–RNA complex formation.
Tofurther study the specificity of RC3H1 interaction to its
A20stem–loop hairpin, we performed EMSA using stem–loop hairpinor
variants thereof (Fig. 5d). A single-nucleotide substitution inthe
loop region virtually did not affect binding; however,substitution
of three nucleotides resulted in slight reduction ofbinding (Fig.
5d). In contrast, a control sequence, which wasgenerated by
concatenating three 7mers underrepresented in our7mer analysis, did
not bind to RC3H1, indicating a specificity forthe A20 hairpin
(Fig. 5d). Moreover, the antisense LNAoligonucleotide, which was
used to modulate A20 mRNA stability(Fig. 4e) and hybridizes to the
loop and the 30part of the stem,
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367
6 NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
reduced the binding of RC3H1-N2 to the 37-nucleotide A20target
sequence, suggesting that the specific blockade of both loopand
stem pronouncedly reduced RC3H1 binding to A20 30UTR(Fig. 5e).
RC3H1 represses A20 and modulates the NF-jB pathway.Notably,
RC3H1-dependent mRNA regulation affects severalNF-kB pathway
regulators5,9. A recent transcriptome-wideRIP-seq study determining
RC3H1-bound transcripts in mousemacrophages revealed the cytokine
TNFa—a typical activator of
NF-kB35,36—and two members of the IkB family, IkBNS(NFKBID) and
IkB-zeta (NFKBIZ)9. In addition to theubiquitin-editing enzyme A20,
we identified IkBa in ourPAR-CLIP as an additional RC3H1 target
transcript (Fig. 6a,b).To examine whether RC3H1 acts in a
pathway-specificmanner, we used previously published mRNA
expression dataof TNFa-treated HEK 293 cells37 to correlate
RC3H1-boundmRNAs with TNFa-induced transcripts. Interestingly,
RC3H1target transcripts showed a greater increase in expression
uponTNFa induction than non-targets, suggesting that RC3H1
actspost-transcriptionally on NF-kB target genes (Fig. 6a and
Log2 (expression level)untreated
4 6 8 10 12 14
4
6
8
10
12
14
Log2
(ex
pres
sion
leve
l)af
ter
DN
A d
amag
e
A20
0 4 9 24 0 4 9 24 0 4 9 24 0 4 9 24
WT
RC3H1 induced
A20 GAPDH
0
5
10
15
20
25
30
Rel
ativ
e m
RN
A le
vel
50
60
70
80
90
100
Per
cent
age
of A
20 m
RN
A le
vel
40
h420h
0.0
0.2
0.4
0.6
0.8
1.0
CD
F
−1.0 −0.5 0.0 0.5 1.0
Log2-fold change (DNA damage/WT)
RC3H1 targets Non-targets
RC3H1 induced WT
h420
70
80
90
100
A20 LNACtr LNA
P < 2.2e–16
Per
cent
age
of A
20 m
RN
A le
vel
Figure 4 | RC3H1 target transcripts are enriched for mRNAs
induced upon DNA damage, and RC3H1 negatively regulates A20 at
the
post-transcriptional level. (a) A scatter plot of mRNA
expression levels of untreated cells and cells treated for 4 h with
200 ng ml� 1 of neocarzinostatin(NCS). The data was retrieved from
Elkon et al.34. RC3H1 30UTR target transcripts are shown in red and
non-targets are shown in black. Among the
RC3H1 targets, A20 was the most differentially expressed mRNA
upon DNA damage. (b) A cumulative distribution function (CDF) plot
of log2-fold changes
upon DNA damage is shown for RC3H1 30UTR targets in red and for
non-targets in black (P value o2.2e-16, Wilcoxon-rank sum test).
(c) RC3H1 induction bydoxycycline treatment specifically leads to
reduced expression of A20 at each time point. mRNA expression level
of A20 and GAPDH (negative control)
are measured by qPCR at 0, 4, 9 and 24 h post-DNA damage induced
by 250 ng ml� 1 of NCS. Averages and s.d.’s (error bar) from three
technical replicatesare shown. A representative data set out of two
independent biological replicates is shown. WT, wild type. (d)
Induction of RC3H1 leads to increased A20
mRNA decay. At 4 h post-DNA damage induced by NCS (250 ng ml�
1), transcription was blocked with actinomycin D, and mRNA
expression levelswere measured by qRT–PCR. Percentage of A20 mRNA
amount at each time point relative to starting point is shown.
Error bars indicate s.d.’s calculated from
three replicates. (e) Transfection of antisense LNA
oligonucleotide targeting the stem–loop structure in HEK293 cells
leads to decreased A20 mRNA decay
(red) in comparison with control (Ctr) LNA transfection (black).
A representative data from two independent experiments are
shown.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367 ARTICLE
NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications 7
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
Supplementary Fig. 5b). Furthermore, RC3H1 induction caused
areduction of basal and stimulus-dependent A20 mRNA andprotein
levels (Fig. 6b,c), similar to A20 expression changesobserved
during the response after DNA damage (Fig. 4c).
Next, we asked whether RC3H1 could modulate NF-kBactivity. NF-kB
activation is mediated via the IKK complex,which catalyses the
phosphorylation of IkB and NF-kB proteins,as well as of other
substrates35,36,38,39. Signalling involves
ICOS (13 nt)
A20 (21 nt)
A20 (37 nt)
A20
10 kbRC3H1-binding site
T AA C T A T A T A A A TT A CCC T T A C A T T A T G T A T GG A G
G A T T T T T T T A A A T T A T A T T G A A A T
Vertebrateconservation
1 _0 _
RC3H1-binding site
0
100
200
300
T -
> C
tran
sitio
n
( ( ( ( ( ( ) ) ) ) ) )
AU
RC3H1-N2 (nM)
––
RC3H1-N1 (nM)
RING ROQ CCCH-Znf PRD
RC3H1
RING ROQ ROQ CCCH-Znf
WT
–
Mut 1
–
Mut 2
– –
Mut 3
RC3H1-N2 (nM)
RC3H1-N2 (nM) RC3H1-N2 (nM)
RC3H1-N2 (nM)
Stem
StemLoopCCCUUACAUUAUGUAUGAGGGCCCUUACAUAAUGUAUGAGGGCCCUUACAAAAAGUAUGAGGGCCUCCCCCCCUGCCCCCCAGC
WTMut1Mut2Mut3
A20 LNA Ctr LNA
h420
40
60
80
100
A20 reporter in siRC3H1/2 A20 reporter in untreated
Empty reporter in untreated
RING
1,60
080
040
020
010
0502512
.501,
600
800
400
200
100
502512.5
0
1,60
080
040
020
010
0502512
.501,
600
800
400
200
100
502512.5
0
1,60
080
040
020
010
0502512
.501,
600
800
400
200
100
502512.5
0
A20 (37 nt)+
RC3H1-N2
Figure 5 | RC3H1 binds to a composite structure-sequence motif
in the A20 30UTR mediated by the CCCH-type Zn-finger domain. (a)
Illustration of the
RC3H1-binding site in the A20 30UTR. The binding sites of RC3H1
in the 30UTR of A20 is shown in red and zoomed in below. T to C
transitions for indicated
base positions are shown. Bases shown in red are forming a
potential stem. Phastcon vertebrate conservation is shown in green.
RC3H1-binding site in the
A20 30UTR contains a stem–loop structure flanked by AU-rich
sequences. (b) The effect of A20 AU-rich element (ARE)-stem–loop
hairpin (37 nucleotide
(nt)) was assayed by transiently transfecting HEK293 cells with
the d2GFP reporter plasmid, which contains the 37-nt sequence
inserted into the 30UTR of
d2GFP. mRNA decay of the reporter transcripts were measured in
mock and RC3H1/RC3H2 knockdown cells. Average and s.d.’s (error
bar) from three
technical replicates are shown. (c) EMSA experiments to examine
the binding mode of RC3H1 to the A20 target site. Increasing
concentration of
recombinant RC3H1-N1 (aa 2–399) or RC3H1-N2 containing an
additional CCCH-type Zn-finger domain (aa 2–452) was incubated with
radiolabelled ICOS
(13 nt), A20 stem–loop (21 nt) and A20 ARE-stem–loop (37 nt),
and free RNA was separated from RNA–protein complexes by native
PAGE. (d) EMSA
experiments to examine the sequence specificity of the A20
stem–loop hairpin. Increasing concentration of recombinant RC3H1-N2
was incubated with
radiolabelled wild-type (WT) A20 stem–loop (21 nt), mutated A20
sequences (Mut 1 and Mut 2) as indicated below, or 21 nt control
sequence (Mut 3)
generated by concatenating three 7mers underrepresented in our
7mer analysis. Mutation in the loop slightly reduces the binding,
and the control
sequence does not virtually bind to RC3H1-N2. (e) Increasing
concentration of antisense LNA oligonucleotide targeting the A20
stem–loop structure
impairs the interaction of RC3H1-N2 and 37 nt ARE-stem–loop.
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367
8 NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
ubiquitin-mediated complex formation of pathway componentsand is
controlled at various levels by negative feedbackmechanisms,
including ubiquitin-editing enzymes such as A20(ref. 40).
Indeed, RC3H1-mediated decrease of A20 protein levels led toa
significant increase of IKK activation (Fig. 6c), which
wasrepressed by additionally expressing exogenous A20 without30UTR
(Supplementary Fig. 5c). In line with this, elevated
Ser536phosphorylation of the IKK substrate p65 was observed (Fig.
6c).However, IkBa degradation and NF-kB DNA-binding activitywere
not detectably affected (Fig. 6c,d). To investigate the impactof
RC3H1 ectopic expression on the kinetics of the NF-kBpathway and to
understand the differential effect on IKK versusNF-kB, a
mathematical modelling approach was used, including
main processes of canonical IKK/NF-kB signalling(Supplementary
Fig. 5d). The model parameters were estimatedbased on the western
blot analysis for A20, IkBa andphosphorylated IKK (Fig. 6c), as
well as qPCR data for IkBaand A20 mRNA levels (Fig. 6b; for details
see Methods section).The model simulations (Supplementary Fig. 5d)
confirmed theexperimental findings that RC3H1 reduces A20 mRNA
andsubsequently A20 protein expression, and, due to the
attenuatedIKK inhibition by A20, leads to an increased IKK
activity. In turn,NF-kB activity is slightly increased by RC3H1,
resulting in anincreased mRNA synthesis of the feedback regulators
IkBa andA20. According to the model, enhanced mRNA syntheses
arecounteracted by the increased mRNA decay mediated by
RC3H1induction. For IkBa, this establishes a compensatory
mechanism
WT (–Dox)RC3H1 induced (+Dox)
0
2
4
6
8
10
Rel
ativ
e m
RN
A le
vel
A20Log2 (expression level)untreated
6 8 10 12
6
8
10
12
Log2
(ex
pres
sion
leve
l)af
ter T
NFa
trea
tmen
t
A20
NFKBIZ
HA
A20
P-IKK
IκBα
P-p65
p65
10 1030 3060 6090 90120 120 TNFα (min)
RC3H1 inducedWT
130 -95 -
95 -
95 -
43 -
70 -
70 -
30 120 30 120 TNFα (min)
NF-κB
Non-specific
TNFα (min)
NF-κB
Non-specific
RC3H1 inducedWT
RC3H1
A20
P-IKK
130 -
70 -
70 -
100 -
100 -
p65
P-p65
*
30 60 30 60
siRC3H1/2Mock
siRC3H1/2Mock
10 30 60 10 30 60 TNFα (min)(kDa)(kDa) 00
00
0 000
0 30 60 120
240 0 30 60 120
240 0 30 60 120
240 0 30 60 120
240
IκBα
TNFα(min)
IκBα
IKKα
Figure 6 | RC3H1 modulates the activation of IKK by TNFa. (a) A
scatter plot of mRNA expression levels of untreated cells and cells
treated for 4 h with10 ng ml� 1 of TNFa. The data were retrieved
from Grimley et al.37. RC3H1 30UTR target transcripts are shown in
red and non-targets are shown in black.Several TNFa-induced mRNAs,
such as A20, IkBa and NFKBIZ, are targets of RC3H1. Among the RC3H1
target transcripts, A20 was the most differentiallyexpressed mRNA
upon TNFa treatment. (b) RC3H1 induction leads to slightly reduced
expression of A20 and IkBa at each time point. mRNA
expressionlevels of A20 (left) and IkBa (right) were measured by
qPCR at indicated time points after TNFa treatment. Representative
data from two independentexperiments are shown. Average and s.e.m.
(error bar) are from three technical replicates. (c) Western blot
analyses of the NF-kB pathway proteins afterTNFa stimulation in
cells with doxycyline (Dox)-dependent RC3H1 expression. HEK293
cells were treated with Dox (1 mg/ml for 72 h), to induceHA-RC3H1.
Subsequently, cells were treated with TNFa as indicated, and
analysed by western blot with the indicated antibodies. RC3H1
upregulationresults in decreased A20 expression, leading to
increased IKK activation (T-loop phosphorylation, P-IKK) and
phosphorylation of p65 (P-p65).
Representative data from two independent experiments are shown.
(d) EMSA analysis of whole-cell extracts for TNFa-induced NF-kB
activity. Cells weretreated as in c. (e) Western blot analyses of
the NF-kB pathway proteins after TNFa stimulation in mock or
RC3H1/2 siRNA-treated HEK293. Cells weretreated with TNFa, and
analysed by western blot with the indicated antibodies. RC3H1
downregulation results in mildly increased A20 expression,
leadingto decreased IKK activation and phosphorylation of p65.
Representative data from two independent experiments are shown. ‘*’
indicates phosphorylated
form of IKK. (f) EMSA analysis of whole-cell extracts for
TNFa-induced NF-kB activity. Cells were treated as in e. Knockdown
of RC3H1 expression reducedthe NF-kB activity.
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367 ARTICLE
NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications 9
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
resulting in largely unaffected IkBa mRNA and protein levelsupon
ectopic RC3H1 expression (Fig. 6b,c).
In contrast, knockdown of RC3H1 and RC3H2 in HEK293cells
resulted in a small, but reproducible, upregulation of A20protein
expression (Fig. 6e), which resulted in decreasedphosphorylation of
IKK (Fig. 6e), decreased phosphorylation ofits substrate p65 (Fig.
6e) and reduced NF-kB DNA-bindingactivity (Fig. 6f). These
observations are also reproduced by themathematical model
(Supplementary Fig. 5d) showing that astrengthening of the A20
feedback leads to changes in activatedlevels of IKK and NF-kB.
Taken together, we could demonstrate that RC3H1 regulatesthe
expression of several NF-kB pathway regulators, therebymodulating
IKK and NF-kB activity.
DiscussionIn the present study, we identified transcriptome-wide
RNA-binding sites of human RC3H1 at nucleotide resolution inHEK293
cells using PAR-CLIP. Our bioinformatic analyses didnot reveal a
well-defined motif as observed for sets of RBPs41;however,
indicated several classes of sequence-structure bindingelements
with U-rich sequences frequently embedded in RNAstem–loop
structures in 30UTRs of target transcripts. Surprisingly,the CDE
core consensus motif (UCYRYGA) deduced byLeppek et al.9 was present
only in a minor fraction of identifiedRC3H1-binding sites. Our
RC3H1 PAR-CLIP data are also inagreement with the concept of a
relaxed CDE revealed bystructural and mutational analyses10, which
indicated a shape-specific rather than sequence-specific
recognition of CDEhairpins by the ROQ domain.
Interestingly, our finding of a PAR-CLIP cluster in the A2030UTR
indicated a yet unrecognized RC3H1-binding mode andspecificity. In
contrast to a typical CDE stem–loop motif, which issufficiently
bound by the ROQ domain, we provide evidence thatthe CCCH-type
Zn-finger domain is involved in contacting theA20 site. A RC3H1
variant containing the CCCH-type Zn-fingerdomain bound with higher
affinity to a non-CDE-like stem–loopstructure with an additional
AU-rich sequence upstream of thehairpin than to the hairpin alone.
In contrast, the N-terminalRC3H1 variant lacking the CCCH-type
Zn-finger domain poorlybound to both of these RNA substrates. The
makeup of RC3H1by distinct RNA-binding domains might allow the
protein torecognize a wider range of RNA structure-sequence
elementsand could function on a larger set of regulatory
elementsthan previously anticipated. The ratio of sequence
andstructure specificity features, determining the strength of
theRC3H1–mRNA association, and the RNA recognition elementfrequency
would influence the regulatory capacity of the RBP.
In addition, our results indicate that RC3H1 interacts with
theCCR4-CAF1-NOT deadenylation complex, and mediates
desta-bilization of RC3H1 target transcripts. RC3H1-bound mRNAsare
encoded by genes with various biological functions outside
ofimmune-response pathways, which is in accordance with themouse
phenotype of Rc3h1 null-knockout that showed perinatallethality
with broad physiological complications42. EnrichedKEGG pathways
included cell cycle, p53 signalling and tumourpathways. By
intersecting our PAR-CLIP target mRNAs withmRNA expression data, we
found that RC3H1 targets areenriched for mRNAs induced by DNA
damage34 and TNFa37.As shown for one of the top mRNA targets, A20,
we postulatethat RC3H1, in general, is involved in fine-tuning or
clearance oftranscriptionally induced mRNAs by shortening their
half-lives.
Our discovery of RC3H1 binding to A20 mRNA and otherTNFa-induced
transcripts prompted us to examine and modelthe impact of RC3H1 on
the IKK/NF-kB pathway. Knockdown of
RC3H1 and RC3H2 increased the expression of A20
proteinexpression, resulting in reduced IKK activity and NF-kB
DNA-binding activity. Vice versa, we show that induction of
RC3H1results in pronounced increase of IKK phosphorylation.
Takentogether, RC3H1 targets several components of the
NF-kBsignalling pathway, and thereby modulates IKK and
NF-kBactivity. The net impact of alterations in RC3H protein
activity onIKK and NF-kB activity in different cell types most
likely dependson various additional (cell-type) specific
parameters. Notably,IKK does not only regulate NF-kB activation but
is also engagedin crosstalk with other pathways40.
The Zn-finger protein A20 is an important negative regulatorof
inflammation21, and several studies have highlighted theclinical
and biological importance of A20. Walle et al.43 recentlyshowed
that negative regulation of the NLRP3 inflammasome byA20 protects
against arthritis. Since RC3H1 is a negative regulatorof A20,
targeting of the RC3H1-A20 mRNA interaction by usingantisense
technologies and concomitant upregulation of A20protein might have
beneficial outcomes in certain diseasescenarios.
In summary, we identified comprehensive RC3H1-bindingsites by
PAR-CLIP, revealing a large number of novel mRNAtargets as well as
novel RC3H1 cis-acting recognition element inthe A20 30UTR. Our
study highlights the importance of post-transcriptional regulation
of gene expression to control crucialcellular signal transduction
pathways.
MethodsAntibodies. For western blots, the following antibodies
were used after dilution to0.5–1 mg ml� 1: anti-HA.11 (Covance,
16B12), anti-FLAG (Sigma, F1804), anti-myc (Sigma, 9E10)
anti-gH2AX(Upstate, JBW301), anti-vinculin (Sigma,
hVIN-1),anti-RC3H1 (Novus, NB100–655), anti-A20 (Santa Cruz
Biotechnology, sc-32525),anti-pIKK (Cell Signal Technology, 2,697),
anti-IKKa (BD Pharmingen, 5,56,532),anti-IkBa (Santa Cruz
Biotechnology, sc-371), anti-p65 (Santa Cruz
Biotechnology,sc-8008P), anti-p-p65 (Cell Signaling Technology,
3033) and polyclonal goat anti-mouse or anti-rabbit
immunoglobulins/horseradish peroxidase (Dako).
Oligonucleotides
siRNAs. siRNA 1 for RC3H1: 50-GCUGGGAAAUACAAAGGAA[dT][dT].siRNA
2 for RC3H1: 50-CCAAGAAAUGUGUAGAAGA[dT][dT].RC3H2:
50-GGAAGAAGGUCGUGUAAGA[dT][dT].
qPCR primers. RC3H1 forward: 50-tggacaaccagaaccacaaa-30
;reverse: 50-GCTGATCCATTTGGTACATCAC-30 .
A20 forward: 50-TGCACACTGTGTTTCATCGAG-30;reverse:
50-ACGCTGTGGGACTGACTTTC-30 .RPL18A forward:
50-GGAGAGCACGCCATGAAG;reverse: 50-AAGATTCGCATGCGGTAGAG-30.GAPDH
forward: 50-AGCCACATCGCTCAGACAC-30 ;reverse:
50-GCCCAATACGACCAAATCC-30 .NFKBIA forward:
50-GAGTCAGAGTTCACGGAGTTC-30 ;reverse: 50-CATGTTCTTTCAGCCCCTTTG-30
.d2GFP forward: 50-GAAGCTTAGCCATGGCTTCCC-30 ;reverse:
50-GATGGCCGCATCTACACATTG-30 .
DNA oligos for d2GFP-A20 30UTR reporter. Sense:
50-GGCCTGTACATATATAATATACCCTTACATTATGTATGAGGGATTTT-30 ;antisense:
50-TCGAAAAATCCCTCATACATAATGTAAGGGTATATTATATATGTACA-30.
RNA oligos. ICOS (13 nucleotide): 50-AUUUCUGUGAAAU-30 .A20 (21
nucleotide): 50-CCCUUACAUUAUGUAUGAGGG-30 .A20 (37 nucleotide):
50-AUAUAUAAUAUACCCUUACAUUAUGUAUGAGGGAUUU-30.Mut1 (21 nucleotide):
50-CCCUUACAUAAUGUAUGAGGG-30 .Mut2 (21 nucleotide):
50-CCCUUACAAAAAGUAUGAGGG-30 .Mut3 (21 nucleotide):
50-CCUCCCCCCCUGCCCCCCAGC-30 .
Plasmids. pENTR4 constructs were generated by PCR amplification
of theRC3H1 and QKI5 coding sequences from cDNA followed by
restriction digest andligation into the pENTR4 (Invitrogen)
backbone, which were further recombinedinto the
pFRT/TO/FLAG/HA-DEST destination vector44 using
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367
10 NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
GATEWAY LR recombinase (Invitrogen) according to manufacturer’s
protocol.Expression plasmids for HA-tagged CNOT1 and CNOT8 were
kind gifts fromDr W. Filipowicz. pENTR4 QKI5 was recombined into
pFRT/FLAG/HA-DEST(Addgene ID: 26,360). The d2GFP reporter plasmids
were generated by cloning thed2GFP (Clontech) coding sequence into
pcDNA5/FRT, and synthetic DNAoligonucleotides containing the
A20-binding site were annealed and ligated intothe 30UTR of d2GFP
using the Xho1/Not1 site.
Cell lines and culture conditions. Flp-In 293 T-REx cells
(Invitrogen) were grownin Dulbecco’s modified Eagle’s medium high
glucose with 10% (v/v) fetal bovineserum, 2 mM L-glutamine. Cell
lines stably expressing FLAG/HA-tagged RC3H1protein were generated
by co-transfection of pFRT/TO/FLAG/HA constructs withpOG44
(Invitrogen). Cells were selected by adding 15 mg ml� 1 blasticidin
and100mg ml� 1 hygromycin (Invivogen). Expression of epitope-tagged
proteins wasinduced by addition of 1 mg ml� 1 doxycyclin. The
expression of FLAG/HA-taggedRC3H1 protein was assessed by western
analysis using mouse anti-HA.11 mono-clonal antibody (Covance). For
quantitative proteomics, cells were grown in SILACmedium as
described before30,45. Briefly, Dulbecco’s modified Eagle’s
mediumGlutaMAX lacking arginine and lysine (PAA) supplemented with
10% dialysedfetal bovine serum (Gibco) was used. Amino acids (84 mg
l� 1 13C615N4 L-arginineplus 146 mg l� 1 13C615N2 L-lysine or 84 mg
l-1 13C6-L-arginine plus 146 mg l� 1 D4-L-lysine) or the
corresponding non-labelled amino acids (Sigma) were added toobtain
‘heavy’ ‘medium’ or ‘light’ cell culture medium, respectively.
Labelled aminoacids were purchased from Sigma Isotec.
Western blot analysis. Total cell lysates were prepared in 1�
SDS–PAGE sampleloading buffer (50 mM Tris (pH 7.5),
mercaptoethanol, 1% SDS, 0.01% bromo-phenol blue and 10% glycerol)
and resolved by SDS–PAGE. Proteins were trans-ferred to
nitrocellulose membrane (Whatman) using a semi-dry blotting
apparatus(Bio-Rad) at constant 20 V for 1 h. The membrane was
blocked in 5% non-fat milkand incubated with primary antibody.
Following incubation for 1 h at roomtemperature, membranes were
washed three times in TBST (150 mM NaCl, 20 mMTris-HCl (pH 7.5) and
0.1 % Tween) and incubated with horseradish peroxidase-conjugated
secondary antibody for 1 h. Following three additional TBST
washes,protein bands were visualized using ECL detection reagent
(GE Healthcare) and aLAS-4000 imaging system (GE Healthcare).
Uncropped images of data shown inthe figures are provided in
Supplementary Fig. 6.
PAR-CLIP. Stably transfected and inducible FLAG/HA-RC3H1
expressing cellswere labelled with 100 mM 4SU or 6SG for 8–9 h.
After labelling the cells, PAR-CLIP was performed essentially as
described in ref. 19. Briefly, for 4SU-2 and one6SG,
ultraviolet-irradiated cells were lysed in NP40 lysis buffer (50 mM
HEPES-KOH at pH 7.4, 150 mM KCl, 2 mM EDTA, 0.5% (v/v) NP40, 0.5 mM
dithiothreitol(DTT) and complete EDTA-free protease inhibitor
cocktail). After mild treatmentwith RNase T1 (Fermentas) at final
concentration of 1 U ml� 1 for 15 min at roomtemperature,
immunoprecipitation was carried out with protein-G magnetic
beads(Invitrogen) coupled to anti-FLAG M2 antibody (Sigma) from
extracts of FLAG/HA-RC3H1 expressing and 4SU-labelled HEK 293 cells
for 1 h at 4 �C. For 4SU-1,a high-salt lysis buffer (50 mM
Tris-HCl, 500 mM NaCl, 1% (w/v) NP40, 1 mM DTTand complete
EDTA-free protease inhibitor cocktail) was used for cell
lysisfollowed by sonication. After mild treatment with RNaseT1 at
final concentrationof 1 unit per ml for 15 min at room temperature,
purification of the RC3H1/RNAcomplex was performed with Flag
magnetic beads (Sigma). Following additionaldigestion by RNase T1
(Fermentas) at final concentration of 10 unit per ml for2 min at
room temperature, beads were incubated with calf intestinal
phosphatase(NEB) and RNA fragments were radioactively end labelled
using T4 polynucleotidekinase (Fermentas). The crosslinked
protein–RNA complexes were resolved on a4–12% NuPAGE gel
(Invitrogen). The SDS–PAGE gel was transferred to a nitro-cellulose
membrane (Whatman) and the protein–RNA complex migrating at
anexpected molecular weight was excised. RNA was isolated by
proteinase K (Roche)treatment and phenol–chloroform extraction,
ligated to 30 adapter. (50-AppTCGTATGCCGTCTTCTGCTTG-InvdT-30 for
4SU-1, 50-AppTCTCGTATCGTATGCCGTCTTCTGCTTG-InvdT-30 for 4SU-2 and
50-AppTCTCTGCTCGTATGCCGTCTTCTGCTTG-InvdT-30 for 6SG), and 50
adapter
(50-rGrUrUrCrArGrArGrUrUrCrUrArCrArGrUrCrCrGrArCrGrArUrC-30),
reverse transcribed and PCRamplified. The amplified cDNA was
sequenced on a HighSeq2000 (Illumina) with a1� 51 nucleotide cycle
for 4SU-2 and 6SG and on a Genome Analyzer II with a1� 36 cycle for
4SU-1.
PAR-CLIP data processing. The PAR-CLIP cDNA sequencing data
wereprocessed by an automated PAR-CLIP analysis pipeline developed
byLebedeva et al.22 with default settings. Briefly, the PAR-CLIP
analysis pipelineperforms the following steps: (1) reads were
mapped to the genome, and uniquelyaligned reads with up to one
mismatch, insertion or deletion were used to buildbinding clusters;
(2) each binding cluster was assigned a score based on the numberof
T to C or G to A mismatches and on the heterogeneity of distinct
readscontributing to the cluster. As the reads should originate
from RC3H1-boundtranscripts, we regarded clusters aligning
antisense to the annotated direction oftranscription as false
positives. (3) Filtering was performed to obtain RC3H1 clusters
sites at an estimated 5% false-positive rate based on the
assigned score for eachcluster. For each cluster, the position with
the highest number of diagnostictransition events was determined,
and we defined this position as the preferredcrosslink site. To
define the consensus clusters, we pooled reads from all
threeexperiments while ensuring that transition events are counted
appropriately (T to Conly in reads originating from 4SU experiments
and G to A only in reads from the6SG experiment). Before the cutoff
determination, clusters had to pass an additionalconsensus filter,
demanding that reads from at least two out of the three
experimentssupport the cluster. The resulting sets of clusters were
denoted as the ‘consensus’ set.Read alignment statistics, cluster
length distribution, target gene identification,cluster
distribution, cluster coverage profiles, conservation profile and
miRNA targetscan were generated by the PAR-CLIP analysis pipeline.
The IGF2BP1 PAR-CLIPdata were obtained from GEO (GSE21578; ref. 19)
and analysed using the samePAR-CLIP analysis pipeline with similar
settings. KEGG pathway and GO termenrichment analysis was performed
using the on-line DAVID programme46,47. Thetop 1,000 transcripts
(ranked by the number of PAR-CLIP diagnostic mutationsfalling into
30UTR) were used for pathway enrichment analysis.
Motif analysis. 7mer occurrences were counted in 41 nucleotide
windows aroundthe crosslink site identified in the 4SU and 6SG
PAR-CLIP experiments usingcustom Perl scripts. To examine the
enrichment of each 7mer motif, 7mer fre-quency occurring in RC3H1
consensus 30UTR-binding sites was compared withthat occurring in
all 30UTR sequences retrieved from UTRdb48. The longest
30UTRsequence for each gene was used in this analysis. To test
whether RC3H1-bindingsites showed a preferred secondary structure,
we used the library routines from theVienna RNA package 1.8.2 (ref.
49) to compute base pairing probabilities within 41nucleotide
sequences centred on the preferred crosslink positions of
30UTR-binding sites. The resulting profiles were accumulated and
averaged over all 30UTRconsensus binding sites or the negative
control 41 nucleotide sequences randomlyselected from the 30UTRs of
RC3H1 target transcripts. Stem–loop enrichmentanalysis was done
based on the output from RNAfold programme in the ViennaRNA
package. For the clustering of structured motifs, we started with
the top 100(ranked by number of diagnostic transition events
divided by expression value)RC3H1-binding sites (41 nucleotide
length sequence centred around the preferredcrosslink site)
determined by PAR-CLIP, and detected initial clusters by
LocARNA(version 1.7.16; refs 26,27) together with RNAclust (version
1.3; ref. 27) to producea hierarchical tree. RNAclust (which uses
LocARNA) was run using defaultparameters together with the RNAsoup
option (./RNAclust.pl –fastatop100_sequences.fasta –dir output_dir/
--RNAsoup). For each subcluster, we usedCMfinder (version 0.2; ref.
28) to search for a subset of sequences that has aconserved
sequence-structure motif. CMfinder generates both a
sequence-structurealignment (called seed alignment) as well as a
covariance model, which we used tosearch for further sequences in
the top 1,000 binding sites for remote members ofthis motif using
cmsearch from the Infernal package (version 1.0.7; ref. 29). Wethen
cut the hierarchical tree at the point where we got structured
motifs with thelargest coverage in the list of top 1,000 binding
sites, which resulted in threestructured motifs. We kept two motifs
(motifs 1 and 2) and discarded the third one,since its seed
alignment consisted of only five entries.
siRNA knockdown and pSILAC. Flp-In 293 T-REx cells were grown in
SILACmedium supplemented with ‘light’ labelled amino acids before
siRNA knockdownexperiments. siRNAs were transfected at a final
concentration of 50 nM usingLipofectamine RNAiMAX (Invitrogen).
Controls (mock) were treated with trans-fection reagent only.
Following 24 h of incubation, siRNA-transfected cells wereswitched
to ‘medium’ labelled SILAC medium, whereas mock control cells
wereswitched to ‘heavy’ labelled SILAC medium. After 24 h of
labelling, cells wereharvested and equal amounts of siRNA- and
mock-transfected cells were pooled,lysed in urea buffer (8 M urea
and 100 mM Tris-HCl, pH 8.3) and sonicated for 20 s(two pulses, 60%
power). Cell debris was removed by centrifugation (14,000g,5 min).
Protein concentration was then measured by the Bradford
colorimetricassay. An amount of 100 mg of proteins were reduced in
2 mM DTT for 30 min at25 �C, and successively free cysteines were
alkylated in 11 mM iodoacetamide for20 min at room temperature in
the dark. LysC digestion was performed by addingLysC (Wako) in a
ratio 1:40 (w/w) to the sample and incubating it for 18 h
undergentle shaking at 30 �C. After LysC digestion, the samples
were diluted three timeswith 50 mM ammonium bicarbonate solution, 7
ml of immobilized trypsin (AppliedBiosystems) were added and
samples were incubated 4 h under rotation at 30 �C.Digestion was
stopped by acidification with 10 ml of trifluoroacetic acid and
trypsinbeads were removed by centrifugation. Fifteen micrograms of
digest were desaltedon STAGE Tips, dried and reconstituted to 20 ml
of 0.5 % acetic acid in water50.A volume of 5 ml of each sample
were injected in duplicate on a Liquidchromatography-tandem mass
spectrometry (LS-MS/MS) system (nanoLC-Ultra1D (Eksigent) coupled
to LTQ-Orbitrap Velos (Thermo)), using a 240-mingradient ranging
from 5 to 40% of solvent B (80% acetonitrile, 0.1 % formic
acid;solvent A¼ 5% acetonitrile and 0.1 % formic acid). For the
chromatographicseparation, B25-cm-long capillary (75 mm inner
diameter) was packed with 1.8 mmC18 beads (Reprosil-AQ, Dr Maisch).
The capillary nanospray tip was generatedusing a laser puller
(P-2000 Laser Based Micropipette Puller, Sutter
Instruments),allowing fritless packing. The nanospray source was
operated with spay voltage of2.1 kV and ion transfer tube
temperature of 260 �C. Data were acquired in data-
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367 ARTICLE
NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications 11
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
dependent mode, with one survey mass spectrometry (MS) scan in
the Orbitrapmass analyser (resolution 60,000 at m/z 400) followed
by up to 20 MS/MS in theion trap on the most intense ions
(intensity threshold¼ 750 counts). Once selectedfor fragmentation,
ions were excluded from further selection for 30 s in order
toincrease new sequencing events. Raw data were analysed using the
MaxQuantproteomics pipeline (v1.3.0.5) and the built-in Andromeda
search engine51
with the International Protein Index Human version 3.71
database.Carbamidomethylation of cysteines was chosen as fixed
modification, andoxidation of methionine and acetylation of N
terminus were chosen as variablemodifications. The search engine
peptide assignments were filtered at 1% FalseDiscovery Rate (FDR)
and the feature match between runs was enabled; otherparameters
were left as default. For SILAC analysis, two ratio counts were set
asthreshold for quantification
RC3H1 protein interactome. For the identification of proteins
directly interactingwith RC3H1, cells are grown in medium
supplemented with either light or heavystable isotope-labelled
amino acids. In the forward experiments, FLAG/HA-taggedRC3H1 was
expressed only in cells cultured in light medium, and in the
reverseexperiments the labelling was swapped. Equal amounts of
cells were mixed and lysedin three pellet volumes of NP40 lysis
buffer (50 mM Tris-HCl (pH 7.5), 150 mM KCl,2 mM EDTA (pH 8.0),
0.5% NP40, 1 mM NaF, 0.5 mM DTT and protease inhibitorcocktail).
The extracts were treated with 1 unit per ml RNase T1 for 5 min at
22 �C tofacilitate the immunoprecipitation and incubated with FLAG
magnetic beads (Sigma;50ml 1� 1 ml cell lysate) for 1 h at 4 �C.
Beads were washed once with NP40 lysisbuffer and treated with 50
unit per ml RNase T1 and 0.25 unit per ml RNase I for5 min at 37 �C
to disrupt RNA-mediated protein interactions. After washing
thebeads once with FLAG elution buffer (100 mM NaCl, 20 mM Tris-HCl
(pH 7.5), 5 mMMgCl2 and 10% glycerol), FLAG/HA-RC3H1 complex was
eluted by adding0.5mg ml� 1 FLAG peptide and rotating for 1 h at 4
�C. mMACS HA magnetic beads(50ml per ml cell lysate) was added to
FLAG eluate and incubated on ice for 30 min.FLAG eluate anti-HA
bead mix was added to the mMACS column after equilibratingwith FLAG
elution buffer. After washing the column three times with 800ml
ice-coldwash buffer I (150 mM NaCl, 50 mM Tris-HCl (pH 7.5), 5%
glycerol and 0.05%NP40) and twice with 500ml ice-cold wash buffer
II (50 mM NaCl, 50 mM Tris-HCl(pH 7.5) and 5% glycerol), a tryptic
digestion was performed on-column byadding 25ml 2 M urea in 100 mM
Tris-HCl (pH 7.5), 1 mM DTT and 150 ng trypsin(Promega). After
in-column digestion for 30 min at room temperature, peptides
wereeluted by adding two times 50ml 2 M urea in 100 mM Tris-HCl (pH
7.5) and 5 mMiodoacetamide. Proteins were further digested
overnight at room temperature beingprotected from light, and
digestion was stopped by adding 1ml trifluoroacetic acid.Resulting
peptides were analysed by mass spectrometry.
Quantitative PCR. Cells were harvested and total RNA was
isolated using Trizol(Invitrogen) according to manufacturer’s
protocol. Total RNA was treated withDNaseI (Invitrogen), and cDNA
synthesis was performed using Superscript III(Invitrogen) with
oligo-dT primer (18–20 nucleotides) or random hexamer
primer(Invitrogen) according to manufacturer’s protocol. qPCR
analysis was performedwith Power SYBR Green PCR Master Mix (ABI)
and ABI light cycler as describedin the manufacturer’s
instructions.
mRNA decay measurement by qRT–PCR. Cells were treated with 5 mg
ml� 1 ofactinomycin D (Sigma-Aldrich) to block the transcription.
At 0, 2 and 4 h post-actinomycin D treatment, total RNA was
harvested using Trizol (Invitrogen)according to manufacturer’s
protocol. Abundance of specific RNA was quantifiedby qRT–PCR. mRNA
levels were normalized against RPL18A mRNA and plottedagainst
time.
LNA transfection. LNA oligonucleotide (Exiqon) antisense to the
RC3H1-boundstem–loop located in the 30UTR of A20
(þAAþATþCCþCTþCAþTAþCAþTAAþT) was transfected at a final
concentration of 100 nM using Lipo-fectamine RNAiMAX (Invitrogen).
For control experiment, control LNA (Exiqon)targeting a region in
the 30UTR of A20 (þTCCAþCCTCþCCCTþCCCþCCþA) not bound by RC3H1 was
transfected as above. Note that þ indicatesthat the following
position is a LNA-modified residue. For mRNA decay assay
afterantisense inhibition at 4 h after the transfection of LNA,
medium was replaced withfresh medium containing 250 ng ml� 1 of
neocarzinostatin (Sigma-Aldrich) toinduce DNA damage and A20
expression. 5 h after induction of DNA damage,mRNA decay assay was
performed. Random hexamer primers (Invitrogen) wereused for cDNA
generation.
mRNA half-life and decay measurements. To calculate the changes
of mRNAdecay rates upon depletion of RC3H1 and RC3H2, cells were
transfected withRC3H1/RC3H2 siRNAs at days 0 and 3. At day 6,
untreated and siRNA-treatedcells were supplemented with actinomycin
D (50 mg ml� 1) to block transcription,and harvested at time 0, 1
and 2 h after actinomycin D treatment. Experiments aredone in
biological replicates for each time point. Total RNA was extracted
usingTrizol, supplemented with 5% Drosophila melanogaster total RNA
and subjected tohigh-throughput sequencing using the TruSeq RNA
sample prep v2 kit (Illumina).
The 12 samples were sequenced on a Illumina HiSeq 2,500 and
mapped to thehuman genome version hg18 using tophat2 (ref. 52).
Non-redundant reads pergene were counted using quasR53. For decay
determination, read counts weredivided by the sum of the reads
matching to the Drosophila transcriptome and bytranscript length,
and log2-transformed. Decay rates were estimated using
linearleast-squares regression on the data of two biological
replicates. We used quantilenormalization of the two sets of decay
rates to remove potential biases fromnormalization. Differences in
decay rates were computed by subtracting decay ratesof
siRNA-treated samples from the decay rates of the mock-treated
samples.4SU-based measurement of mRNA half-lives was performed as
described in ref. 32.Briefly, HEK293 cells were treated with 100 mM
4SU for 60 min. Total cellular RNAwas isolated using Trizol
reagent. Biotinylation of 4SU-labelled RNA wasperformed using
EZ-Link Biotin-HPDP
(N-[6-(biotinamido)hexyl]-30-(20-pyridyldithio)propionamide)
(Pierce) dissolved in dimethylformamide.Biotinylation was carried
out in 10 mM Tris (pH 7.4), 1 mM EDTA and0.2 mg ml� 1 biotin-HPDP
at a final RNA concentration of 100 ng ml� 1 for 1.5 h atroom
temperature. An amount of 50–100 mg of total RNA was used for
thebiotinylation reaction. Unbound Biotin-HPDP was efficiently
removed bychloroform:isoamyl alcohol (24:1) extraction using
Phase-lock-gel (heavy) tubes(Eppendorf). Then, a 1/10 volume of 5 M
NaCl and an equal volume ofisopropanol were added, and RNA was
precipitated at 20,000g for 20 min. Thepellet was washed with an
equal volume of 75% ethanol and precipitated at 20,000gfor 10 min.
The pellet was resuspended in 50–100 ml RNase-free water.
Afterdenaturation of RNA samples at 65 �C for 10 min followed by
rapid cooling on icefor 5 min, biotinylated RNA was captured using
mMACS streptavidin beads andcolumns (Miltenyi). Up to 100mg of
biotinylated RNA were incubated with 100mlof mMACS streptavidin
beads with rotation for 15 min at room temperature. Thebeads were
transferred and magnetically fixed to the columns. Columns
werewashed three times with 1 ml 65 �C washing buffer (100 mM
Tris-HCl (pH 7.4)10 mM EDTA, 1 M NaCl and 0.1% Tween20) followed by
three washes with roomtemperature washing buffer. To recover the
unlabelled pre-existing RNA the flow-through of the first two
washes was collected and combined. Labelled RNA waseluted by adding
100 ml of freshly prepared 100 mM DTT followed by a secondelution 5
min later. RNA was recovered from the washing fractions and
eluatesusing the RNeasy MinElute Spin columns (Qiagen). Total RNA
(1.5 mg) and newlytranscribed RNA (280 ng) were amplified and
labelled using the Affymetrix One-Cycle Target Labeling Kit
according to the manufacturer’s protocol. As newlytranscribed RNA
mainly consists of mRNA, it was amplified and labelled accordingto
the manufacturer’s protocol for mRNA. The amplified and
fragmentedbiotinylated cRNA was hybridized to Affymetrix Human Gene
1.0 ST Arrays usingstandard procedures. Data were processed and
analysed with R and Bioconductor.To calculate RNA half-lives,
CEL-files of all samples from all conditions (includingtotal RNA,
newly transcribed RNA and pre-existing RNA) were normalizedtogether
using the GCRMA algorithm. Only probe sets called ‘present’ in all
threereplicates of all three RNA subsets under study were included
in the analysis oftranscript half-lives. Calculation of RNA
half-lives was done as performed 32.Statistical comparison of
half-life values between groups was performed using theWilcoxon
rank-sum test.
Recombinant protein expression and purification. DNA encoding
the RINGand ROQ domains (RC3H1-N1; aa 2–399) or the RING, ROQ and
CCCH-Znfdomains (RC3H1-N2; aa 2–452) was subcloned into the pQLinkH
vector54. Thegenes were expressed as N-terminal His7-tagged
proteins at 17 �C in Escherichiacoli Rosetta 2 (DE3, (Novagen)
using a LEX ultra-high-throughput bench-topbioreactor (Harbinger
Biotech). Cells were grown at 37 �C in Terrific Brothmedium and
induced at an OD600 of 2.0–2.5 with 0.5 mM isopropyl
b-D-1-thiogalactopyranoside. For purification, cells were
resuspended in phosphate-buffered saline lysis buffer (1�
phosphate-buffered saline, pH 7.4, 0.5 M NaCl, 5%(v/v) glycerol and
0.5 mM DTT), supplemented with 0.25% (w/v) 3-[(3-cholamidopropyl)
dimethylammonio]-1-propanesulfonate, 0.1 mM phenylmethyl
sulfonylfluoride, 1 U ml� 1 RNase-free DNase I (Qiagen) and one
tablet of EDTA-freeComplete Protease Inhibitor (Roche). The
purification procedure comprisesmechanical cell lysis by sonication
(SONOPULS HD 2200, Bandelin), an Ni/Znaffinity chromatography on a
5-ml HisTrap FF crude column (GE Healthcare), anda size-exclusion
chromatography on a Superdex 200 prep grade column (XK26� 60, GE
Healthcare). The His7 tag was cleaved with tobacco etch virus
proteasebefore the gel filtration step, followed by a reapplication
of the cleaved protein onthe Ni/Zn affinity column. The
purification of protein constructs comprising theRING, ROQ and
CCCH-Znf domains additionally included a
cation-exchangechromatography on a Source 30S column (HR 16� 10, GE
Healthcare).
Electrophoretic mobility shift assay. The EMSA was performed
according toRyder et al.55 with the following modifications: RNA
was prepared by 50-endlabelling of commercially synthesized RNA
oligonucleotides with [g-32P]-ATPusing T4 polynucleotide kinase
(NEB). Labelled RNA was gel-purified, eluted andadjusted with H2O
to 1 pmolml� 1. Labelled RNA (50 fmol) was used per 20 mlreaction.
Before binding reactions, a master mix containing labelled RNA,
1�binding buffer (20 mM Tris-HCl (pH 7.5), 50 mM KCl, 5 mM MgCl2,
20 mM ZnSO4and 10% glycerol), 2 mM DTT, 0.05 mg ml� 1 BSA and 5 mg
ml� 1 heparin washeated at 90 �C for 1 min and gradually cooled
down to room temperature. In
ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367
12 NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
parallel, a dilution series of 10� protein stocks was prepared
in 1� proteindilution buffer (1� binding buffer and 5 mg ml� 1
heparin). For each bindingreaction, 2 ml of the 10� protein stock
was added to 18 ml of the mastermix atroom temperature for 2 h.
After addition of 4 ml 6� loading buffer (30% glycerol,bromophenol
blue and xylene cyanol), RNA-protein complexes were resolved
bynondenaturing PAGE (6% polyacrylamide, 0.5� Tris-borate-EDTA
(TBE) and 5%glycerol) in ice-cold 0.5� TBE buffer containing 20mM
ZnSO4 at 100 V for40 min. The protein-bound RNA and the free RNA
were quantified using aphosphorimager.
To determine NF-kB DNA-binding activity, EMSA was performed
according toHinz et al.56 using the following protocol. An amount
of 4–10 mg of whole-celllysate was mixed with radioactive-labelled
(25,000 c.p.m.) oligonucleotidescontaining a NF-kB site
(50-gatcCAGGGCTGGGGATTCCCCATCTCCACAGG-30 and
50-gatcCCTGTGGAGATGGGGAATCCCCAGCCCTG-30), 2 mg poly(dI-dC), 1 mg
BSA and 1 mM DTT in 20ml reaction buffer (20 mM HEPES (pH 8.4),60
mM KCI and 8% Ficoll) Samples were incubated for 30 min at 25 �C
andanalysed by native PAGE (5 % gels; TBE buffer), followed by
autoradiography.
Microarray data processing. Microarray raw data for DNA damage
response andTNFa response were retrieved from GEO accessions
GSE1676 and GSE28548,respectively. Robust multi-array average
background correction and quantilenormalization was applied using
affyR Bioconductor packages57. For the analysis ofAffymetrix Human
Genome U133 Plus 2.0 Array, probe set intensities mapping tothe
same gene were averaged to summarize into gene intensities, and
genes withlog2 steady-state expression level o5 were filtered
out.
Mathematical modelling of NF-jB pathway. The computational model
of thecanonical IKK/NF-kB system is described by an ordinary
differential equationsystem:
A200 ¼ k9�A20mRNA � k3�A20
A200mRNA ¼ k12�NFkB� k7�A20mRNA�RC3H1A20
IKK0act ¼ TNFa�k10�e�A20þ k2=ðk2 þA20Þ� k4�IKKact
NF-kB j IkBa0 ¼ k1�IkBa�NFkB� k14�NF-kB j IkBa� k6�IKKact�NF-kB
j IkBa
IkBa0 ¼ k14�NF-kB j IkBa� k1�IkBa�NFkBþ k11�IkBamRNA �
k5�IkBa
IkBa0mRNA ¼ k13�NFkB� k8�IkBamRNA�RC3H1IkBa
NFkB0 ¼ k6�IKKact�NF-kB j IkBa� k1�IkBa�NFkBþ k14�NF-kB j
IkBaThe western blot data of three replicates for phosphorylated
IkBa and IKK
upon TNFa stimulation with and without RC3H1 induction were
quantified usingImageJ. The mRNA levels of IkBa and A20 upon TNFa
stimulation with andwithout RC3H1 induction were measured with
qPCR. The parameters of the modelwere estimated with the
Data2Dynamics software package in MATLAB (R2013b,The Mathworks
Inc., Natick, MA) using the build-in function lsqnonlin and
latinhypercube parameter sampling58. One representative parameter
set is given inSupplementary Table 3.
References1. Schoenberg, D.R. & Maquat, L.E. Regulation of
cytoplasmic mRNA decay. Nat.
Rev. Genet. 13, 246–259 (2012).2. Hao, S. & Baltimore, D.
The stability of mRNA influences the temporal order of
the induction of genes encoding inflammatory molecules. Nat.
Immunol. 10,281–288 (2009).
3. Vinuesa, C.G. et al. A RING-type ubiquitin ligase family
member required torepress follicular helper T cells and
autoimmunity. Nature 435, 452–458(2005).
4. Athanasopoulos, V. et al. The ROQUIN family of proteins
localizes tostress granules via the ROQ domain and binds target
mRNAs. FEBS J. 277,2109–2127 (2010).
5. Pratama, A. et al. Roquin-2 shares functions with its paralog
Roquin-1 in therepression of mRNAs controlling T follicular helper
cells and systemicinflammation. Immunity 38, 669–680 (2013).
6. Yu, D. et al. Roquin represses autoimmunity by limiting
inducible T-cell co-stimulator messenger RNA. Nature 450, 299–303
(2007).
7. Glasmacher, E. et al. Roquin binds inducible costimulator
mRNA and effectorsof mRNA decay to induce microRNA-independent
post-transcriptionalrepression. Nat. Immunol. 11, 725–733
(2010).
8. Vogel, K.U. et al. Roquin paralogs 1 and 2 redundantly
repress the Icos andOx40 costimulator mRNAs and control follicular
helper T cell differentiation.Immunity 38, 655–668 (2013).
9. Leppek, K., Schott, J., Reitter, S., Poetz, F., Hammond, M.C.
& Stoecklin, G.Roquin promotes constitutive mRNA decay via a
conserved class of stem-looprecognition motifs. Cell 153, 869–881
(2013).
10. Schlundt, A. et al. Structural basis for RNA recognition in
roquin-mediated post-transcriptional gene regulation. Nat. Struct.
Mol. Biol. 21,671–678 (2014).
11. Tan, D., Zhou, M., Kiledjian, M. & Tong, L. The ROQ
domain of Roquinrecognizes mRNA constitutive-decay element and
double-stranded RNA. Nat.Struct. Mol. Biol. 21, 679–685 (2014).
12. Schuetz, A., Murakawa, Y., Rosenbaum, E., Landthaler, M.
& Heinemann, U.Roquin binding to target mRNAs involves a winged
helix-turn-helix motif. Nat.Commun. 5, 5701 (2014).
13. Maruyama, T. et al. Roquin-2 promotes ubiquitin-mediated
degradation ofASK1 to regulate stress responses. Sci. Signal. 7,
ra8 (2014).
14. Brooks, S.A. & Blackshear, P.J. Tristetraprolin (TTP):
Interactions with mRNAand proteins, and current thoughts on
mechanisms of action. Biochim. Biophys.Acta 1829, 666–679
(2013).
15. Mukherjee, N. et al. Global target mRNA specification and
regulation by theRNA-binding protein ZFP36. Genome Biol. 15, R12
(2014).
16. Shaw, G. & Kamen, R. A conserved AU sequence from the 3’
untranslatedregion of GM-CSF mRNA mediates selective mRNA
degradation. Cell 46,659–667 (1986).
17. Caput, D., Beutler, B., Hartog, K., Thayer, R.,
Brown-Shimer, S. & Cerami, A.Identification of a common
nucleotide sequence in the 3’-untranslated region ofmRNA molecules
specifying inflammatory mediators. Proc. Natl Acad. Sci. USA83,
1670–1674 (1986).
18. Chen, C.Y. & Shyu, A.B. AU-rich elements:
characterization and importance inmRNA degradation. Trends Biochem.
Sci. 20, 465–470 (1995).
19. Hafner, M. et al. Transcriptome-wide identification of
RNA-binding proteinand microRNA target sites by PAR-CLIP. Cell 141,
129–141 (2010).
20. Shembade, N. & Harhaj, E.W. Regulation of NF-kappaB
signaling by the A20deubiquitinase. Cell. Mol. Immunol. 9, 123–130
(2012).
21. Ma, A. & Malynn, B.A. A20: linking a complex regulator
of ubiquitylation toimmunity and human disease. Nat. Rev. Immunol.
12, 774–785 (2012).
22. Lebedeva, S. et al. Transcriptome-wide analysis of
regulatory interactions of theRNA-binding protein HuR. Mol. Cell
43, 340–352 (2011).
23. Ogata, H., Goto, S., Sato, K., Fujibuchi, W., Bono, H. &
Kanehisa, M. KEGG:Kyoto Encyclopedia of Genes and Genomes. Nucleic
Acids Res. 27, 29–34(1999).
24. Ashburner, M. et al. Gene ontology: tool for the unification
of biology. TheGene Ontology Consortium. Nat. Genet. 25, 25–29
(2000).
25. Heyne, S., Costa, F., Rose, D. & Backofen, R.
GraphClust: alignment-freestructural clustering of local RNA
secondary structures. Bioinformatics 28,i224–i232 (2012).
26. Will, S., Joshi, T., Hofacker, I.L., Stadler, P.F. &
Backofen, R. LocARNA-P:accurate boundary prediction and improved
detection of structural RNAs. RNA18, 900–914 (2012).
27. Will, S., Reiche, K., Hofacker, I.L., Stadler, P.F. &
Backofen, R. Inferringnoncoding RNA families and classes by means
of genome-scale structure-basedclustering. PLoS Comput. Biol. 3,
e65 (2007).
28. Yao, Z., Weinberg, Z. & Ruzzo, W.L. CMfinder--a
covariance model basedRNA motif finding algorithm. Bioinformatics
22, 445–452 (2006).
29. Nawrocki, E.P., Kolbe, D.L. & Eddy, S.R. Infernal 1.0:
inference of RNAalignments. Bioinformatics 25, 1335–1337
(2009).
30. Ong, S.E., Foster, L.J. & Mann, M. Mass
spectrometric-based approaches inquantitative proteomics. Methods
29, 124–130 (2003).
31. Doidge, R., Mittal, S., Aslam, A. & Winkler, G.S.
Deadenylation of cytoplasmicmRNA by the mammalian Ccr4-Not complex.
Biochem. Soc. Trans. 40,896–901 (2012).
32. Dolken, L. et al. High-resolution gene expression profiling
for simultaneouskinetic parameter analysis of RNA synthesis and
decay. RNA 14, 1959–1972(2008).
33. Selbach, M., Schwanhausser, B., Thierfelder, N., Fang, Z.,
Khanin, R. &Rajewsky, N. Widespread changes in protein
synthesis induced by microRNAs.Nature 455, 58–63 (2008).
34. Elkon, R., Linhart, C., Sharan, R., Shamir, R. & Shiloh,
Y. Genome-wide in silicoidentification of transcriptional
regulators controlling the cell cycle in humancells. Genome Res.
13, 773–780 (2003).
35. Napetschnig, J. & Wu, H. Molecular basis of NF-kappaB
signaling. Ann. Rev.Biophys. 42, 443–468 (2013).
36. Huang, T.T., Wuerzberger-Davis, S.M., Wu, Z.H. &
Miyamoto, S. Sequentialmodification of NEMO/IKKgamma by SUMO-1 and
ubiquitin mediatesNF-kappaB activation by genotoxic stress. Cell
115, 565–576 (2003).
37. Grimley, R. et al. Over expression of wild type or a
catalytically deadmutant of Sirtuin 6 does not influence NFkappaB
responses. PLoS ONE 7,e39847 (2012).
38. Renner, F. & Schmitz, M.L. Autoregulatory feedback loops
terminating theNF-kappaB response. Trends Biochem. Sci. 34, 128–135
(2009).
39. Hayden, M.S. & Ghosh, S. NF-kappaB, the first
quarter-century:remarkable progress and outstanding questions.
Genes Dev. 26, 203–234(2012).
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8367 ARTICLE
NATURE COMMUNICATIONS | 6:7367 | DOI: 10.1038/ncomms8367 |
www.nature.com/naturecommunications 13
& 2015 Macmillan Publishers Limited. All rights
reserved.
http://www.nature.com/naturecommunications
-
40. Hinz, M. & Scheidereit, C. The IkappaB kinase complex in
NF-kappaBregulation and beyond. EMBO Rep. 15, 46–61 (2014).
41. Ray, D. et al. A compendium of RNA-binding motifs for
decoding generegulation. Nature 499, 172–177 (2013).
42. Bertossi, A. et al. Loss of Roquin induces early death and
immune deregulationbut not auto